Seminar: 1:00-2:30pm EDT
Student Session: 3:00-4:00 EDT
A Comparison of Two Frameworks for Multi-State Modelling, Applied to Outcomes after Hospital Admissions with COVID-19
In this talk, Chris will compare two multi-state modelling frameworks that can be used to represent observed dates of events for a set of individuals. The methods are applied to data from people admitted to hospital with COVID-19 in England, to estimate the probability of admission to ICU, the probability of death in hospital for patients before and after ICU admission, the lengths of stay in hospital, and how these vary with age and gender. One modelling framework is based on defining transition-specific hazard functions for competing risks. A less commonly-used framework defines partially-latent subpopulations who will experience each subsequent event, and uses a mixture model to estimate the probability that an individual will experience each event, and the distribution of the time to the event given that it occurs. Chris will compare the advantages and disadvantages of these two frameworks, in the context of the COVID-19 example. The issues include the interpretation of the model parameters, the computational efficiency of estimating the quantities of interest, implementation in software and assessing goodness of fit. In the example, it is found that some groups appear to be at very low risk of some events, in particular ICU admission, and these are best represented by using “cure-rate” models to define transition-specific hazards. The models described can be implemented in the R package “flexsurv”, which allows arbitrarily-flexible distributions to be used to represent the cause-specific hazards or times to events.
Christopher Jackson is a Senior Investigator Statistician at the MRC Biostatistics Unit, University of Cambridge. His research involves statistical methods in models that combine evidence to evaluate health policies. His published research includes papers on Bayesian evidence synthesis, decision theory, model assessment and comparison, survival analysis, multi-state modelling and longitudinal data. He has also developed several popular R packages, and co-authored “The BUGS Book”.
The student session after the talk will allow students to ask Chris questions about his research, the talk, the recommended paper or career opportunities. If you’re a student, make sure to register for this session.
There are two papers this month (see titles and links below). The paper by Crowther and Lambert gives an accessible introduction to fully-parametric multi-state modelling using cause-specific hazards. In the seminar, Chris will contrast this class of models to a different class of fully-parametric multi-state models based on mixtures, and apply the methods to COVID hospital admissions data. The paper by Ieva et al. describes a similar application of multi-state models to data from hospital admissions, and uses both fully-parametric and semi-parametric methods.
• Crowther, M. J., & Lambert, P. C. (2017). Parametric multistate survival models: flexible modelling allowing transition‐specific distributions with application to estimating clinically useful measures of effect differences. Statistics in Medicine, 36(29), 4719-4742. https://onlinelibrary.wiley.com/doi/full/10.1002/sim.7448
• Francesca Ieva, Christopher H Jackson and Linda D Sharples. Multi-state modelling of repeated hospitalisation and death in patients with heart failure: The use of large administrative databases in clinical epidemiology. SMMR, 2017 Volume: 26 issue: 3, page(s): 1350-1372 (published online 2015). https://doi.org/10.1177/0962280215578777